Toxicology in the 21st Century Program (Tox21) - Computational Toxicology

National Center For Advancing Translational Sciences

Investigators

Linked publications, trials & patents

Abstract

The Tox21 Programs federal partners include the Environmental Protection Agency (EPA), the Food and Drug Administration (FDA) and NIH, with leadership from NCATS and the National Toxicology Program (NTP) at the National Institute of Environmental Health Sciences (NIEHS). These agencies work together to advance in vitro toxicological testing. The Tox21 Program can be separated into three NCATS teams: Systems Toxicology, Genomic Toxicology, and Computational Toxicology. The Tox21 Computational Toxicology team has enhanced a variety of tools that are routinely used by Tox21 partners to access each other’s data. The team performed data analysis of eight assays that were identified, developed, optimized, and/or screened by the Tox21 systems toxicology team and gene expression and high throughput neurotoxicity assay data generated by the Tox21 Genomic Toxicology team. These activities include normalization and correction, fitting of concentration-response curves to generate potency and efficacy measures, classification of curves based on a set of criteria that included significance of fit (measured by p-values), completeness of fit, and efficacy, evaluation of assay performance by data reproducibility, data driven selection of compounds for follow up studies, and identification of genes and pathways involved in cell responses to chemical exposure. In addition, the Tox21 Computational Toxicology team has updated the web-based, automated structure-activity relationship (SAR) analysis tool for the systems toxicology team to conduct SAR analysis on all 10K library screens. The Tox21 Computational Toxicology team has also updated the Tox21 Assay Tracking System that stores the assay annotations and detailed experimental conditions and screening protocols for all the Tox21 assays. The 10K data from all assays screened up to FY24 have been made public in PubChem totaling 268 assay entries (AIDs) and over 122 million data points. The Tox21 public data browser has been updated with the latest assay results from the 10K library screens totaling 90 assays. This browser provides the public with visualization of Tox21 qHTS data including concentration-response curves, curve fitting results and different activity metrics along with chemical structure and analytical QC results. Data are searchable by assay and/or chemical. Results from multiple assays and/or chemicals can be overlaid for ease of comparison. All data as well as assay descriptions and detailed screening protocols (SLPs) are available for download. The Tox21 Computational Toxicology team continued to work with the Tox21 chemical working group to analyze the Tox21 10K library chemical QC results, including the 4-month compound stability test results. A paper documenting this effort was published in Chemical Research in Toxicology in FY25. The Tox21 Computational Toxicology team is also leading one of the Tox21 cross- partner projects (CPP) Expansion of Pathway Coverage by Tox21 High-Throughput Screening Assays for Better Prediction of Adverse Drug Effects, which utilizes the NCATS BioPlanet (https://tripod.nih.gov/bioplanet/) as a tool to define the biological space and identify assays for screening. This project has generated twelve assay datasets up to FY25 and these data have been applied to build computational models to predict drug-induced liver injury (DILI) and cardiotoxicity. This project has resulted in one manuscript published in Environmental Health Perspectives in FY25. In addition, the Tox21 Computational Toxicology team has combined chemical structure information with Tox21 in vitro qHTS assay data to build machine learning models for the prediction of systemic acute in vivo toxicity, resulting in a manuscript published in Toxicology and Appllied Pharmacology in FY25. The Tox21 Computational Toxicology team also led the data analysis and interpretation in multiple other Tox21 CPPs in collaboration with the Tox21 Systems Biology team and other Tox21 partners. For Tox21 CPP #9 - Retrofitting existing Tox21 HTS assays with metabolic capability, the team analyzed the data from Tox21 assay with and without metabolic capacity and found that the cytochrome P450 assay data are highly predictive of compounds that require metabolic activation. By comparing the MSTI assay data from Tox21 CPP #12, Evaluating thiol reactivity using the MSTI Assay protocol, with the compound activity profiles across all Tox21 assays, the team evaluated the utility of the MSTI assay in predicting compound promiscuity and cytotoxicity. These collaborative efforts resulted in two papers published in Toxicology and Applied Pharmacology and Chemical Research in Toxicology, respectively, in FY25. In FY25, the Tox21 Computational Toxicology team has been collaborating with the Informatics Core of DPI and the Collaborative Research, Informatics, and Special Programs Branch to develop machine learning approaches utilizing NCATS in-house biomedical data and Tox21 bioactivity profiles for rare disease drug repurposing, resulting in one publication in PLoS One in FY25. The Tox21 Computational Toxicology team continued to collaborate with the TDB biology team in FY25 to develop computational models for the identification of novel antiviral compounds, including anti-SARS-CoV-2 compounds, using chemical structure and viral sequence information. A manuscript summarizing the results has been published in Communications Chemistry in FY25. In addition, the team identified pathways characteristic of 16 common cancer types and potential drugs for intervention through multi-omics analysis in combination with data from cancer profiling screens conducted by the TDB biology team, resulting in a paper published in the Pharmacogenomics Journal in FY25. The Tox21 Computational Toxicology team collaborated with FDA investigators as part of the Translational Science Interagency Fellowship (TSIF) program to identify potential genetic predispositions for drug-induced liver injury (DILI) through whole exome sequencing analyses. This project produced two publications in FY25, in Liver International and Gastro Hep Advances, respectively. Moreover, in collaboration with NIA investigators, the Tox21 Computational Toxicology team helped to identify potential drug repurposing candidates for C9orf72-related ALS/FTD using large-scale genomic data with supporting evidence from in vitro assay data. The study has been published in Cell Genomics in FY25. Public Health Impact Statement During the course of a lifetime, most people are exposed to many different environmental chemicals. These substances can be found in food, water, household cleaning products and elsewhere. In some cases, these chemicals can be toxic, and in others, researchers lack sufficient data about safety. Medicines also contain chemicals, and in fact, more than 30 percent of promising pharmaceuticals have failed in human clinical trials because they are found to be toxic, despite promising pre-clinical studies in animal and other models. The Tox21 Program works to create alternative methods for assessing chemical toxicity that are less expensive and time-consuming than traditional approaches will improve how scientists evaluate environmental chemicals and develop new medicines to benefit public health.